Are "Climate-Related Deaths" Decreasing?

A popular contrarian trope promoted by Bjorn Lomborg and his followers is that "climate-related deaths" have plummeted since 1920. Recently, he's claimed that there has been a 97.5% reduction in climate-related deaths since 1920.

This is an extremely misleading graph, and it's not too difficult to uncover why. Lomborg got his data for "climate-related" deaths from the Our World in Data website, which lists deaths from "natural disasters" (OWD does not specify which of these deaths from natural disasters are "climate-related." I went to the site to see if I can reproduce Lomborg's graph. He only lists "floods, droughts, storms, and wildfires" but OWD includes other causes of death, including extreme weather and temperature. So below I show all on the OWD site except for earthquakes and volcanic activity. I suspect this is what Lomborg did.

It should be easy to see what Lomborg has done:

1. He relabeled OWD's data for deaths from natural disasters as "climate-related" after removing  deaths from earthquake and volcanic activity from OWD's numbers.
2. He cut off the first 20 years of data to hide the fact that there was an increase in deaths from natural disasters from 1900 to 1920 followed by a decrease.
3. He calculated decadal means from the data to hide the fact that, even during the 1920s and 1930s, there were years with relatively few deaths from natural disasters. There are spikes in deaths from individual events that kill large numbers of people.
4. He lumped all the deaths he labeled "climate-related" together hiding the fact that years with high numbers of deaths come exclusively from floods and droughts.

But the most misleading part of Lomborg's graph is that it ignores the influence of non-climate-related variables that explain years with events that caused large numbers of deaths. For instance:

1. The vast majority of these deaths from natural disasters come from 3 countries: India, China, and Bangladesh. These countries were poor with large populations, meaning they had less infrastructure to respond to disasters that occur.
2. The years with very high death counts are largely explained by the consequences of war and terrible policy decisions, not climate (see below).
3. Societies have more recently built systems to detect natural disasters and warn citizens and developed better infrastructure with weather-specific building codes, etc. This is even true in China. These adaptations have allowed societies to reduce the death toll from natural disasters even as climate-related disasters show signs of increasing.

To illustrate point 2 above, one resource examined what caused those large spikes in deaths from natural disasters from 1920 to 1965. He discovered that virtually all of these years with large amount of deaths were from China and India and occurred as a result of war and/or bad policy decisions. Here's a summary of his findings:

• 1920-1921 - Deaths from crop failures in China as the direct result of the Zhili-Anhui War.
• 1928 - Deaths from policy decisions in China having to do with opium production and ongoing wars which cut into food supplies.
• 1931 - Deaths from flooding in China during the Chinese Civil War. Deaths were due to disease and lack of food following the flood and made worse by Japan's invasion of Manchuria.
• 1938-1939 - Deaths from floods caused by the destruction of dikes by Chinese Nationalist forces to stop Japanese troops from advancing.
• 1942-1943 Famine in India and Bangladesh came from the redirection of food supplies during war.
• 1959 - Flooding in China was made worse by China's "Great Leap Forward" policy decisions.
• 1965 - Drought in India is reported at 1.5 million, but that may not be accurate. It may be as small as 5000 dead.

The OWD correctly labels its data as deaths due to natural disasters, and OWD doesn't include any attribution study to evaluate how many of these deaths were "climate-related." In fact, the managers of the EM-DAT database (from which OWD gained their data) has explicitly stated that it's impossible to make attributions to causation from these numbers. They clearly write, "it is impossible to draw conclusions about the underlying causes of the century-long trend in disaster mortality based on EM-DAT numbers alone.”

Lomborg appears to have just removed volcanism and earthquakes from the EM-DAT data and then assumed all the remaining deaths were "climate-related." This is dishonest. It ignores confounding variables that impact the number of deaths from natural disasters that have nothing to with climate. Clearly, the data shows that years with large numbers of deaths from natural disasters in the 20th century were in predominantly in poor countries with large populations and were associated with warfare and policy decisions, not "climate;" they cannot be summarily lumped in with "climate-related deaths. 

Lomborg obscures these facts by lumping all deaths from natural disasters (excluding volcanism and earthquakes) together and arbitrarily relabeling the remaining deaths as "climate-related" without accounting for any confounding variables or outliers in the data. Then he plotted the data only from 1920 on as decadal means to give the appearance of a near constant decline in deaths from 1920 to the present. This is dishonest.